Meta signs multibillion-dollar deal to use Amazon chips for AI

Meta has signed a multibillion-dollar, multiyear agreement with Amazon to rent Graviton processors for its AI operations, highlighting its aggressive expansion of AI infrastructure despite plans to cut 8,000 jobs to finance it.

The deal gives Meta access to Amazon Web Services’ Graviton5 chip line at scale, starting with tens of millions of Graviton cores at launch, with room to expand as Meta’s AI ambitions grow. The agreement was confirmed by Nafea Bshara, Amazon vice president and co-founder of its Annapurna Labs chips unit.

Meta plans to invest heavily in AI by 2026, spending $115-135 billion on data centres, chips, and AI engineers. To fund this, they are cutting 8,000 jobs starting May 20 and freezing hiring for 6,000 open positions, shifting investment from personnel to AI infrastructure.

Why Meta is betting on Amazon chips and what they actually do

To understand why this deal matters, it helps to understand the difference between the two main types of chips powering the AI industry. Graphics Processing Units (GPUs), the kind made by Nvidia, are essential for training large AI models. But once those models are trained and deployed, much of the work shifts to central processing units (CPUs).

Tasks like real-time reasoning, code generation, search, and orchestrating complex multi-step AI agent workflows are CPU-intensive. This is where Graviton fits in.

The Graviton5 chip is built specifically for this kind of work. It features 192 cores and a cache five times larger than the previous generation, which reduces communication delays between cores by up to 33%.

This enables faster data processing and greater bandwidth, critical for agentic AI systems that need to continuously reason through and execute complex tasks at scale. Graviton5 is also built on 3-nanometer chip technology, delivering up to 25% better performance than its predecessor while consuming less energy.

“As we scale the infrastructure behind Meta’s AI ambitions, diversifying our compute sources is a strategic imperative,” said Santosh Janardhan, head of infrastructure at Meta.

“AWS has been a trusted cloud partner for years, and expanding to Graviton allows us to run the CPU-intensive workloads behind agentic AI with the performance and efficiency we need at our scale.”

Similar read: Meta to cut 8,000 jobs in May as AI reshapes global tech 

Meta has already partnered with Nvidia and Advanced Micro Devices for significant chip agreements and collaborated with Arm Holdings on CPU development. The deal with Amazon expands its chip strategy by adding another partnership.

This diversified approach aims to lessen its reliance on any single provider as the need for AI computing power increases throughout the industry.

The bigger picture: Meta is rebuilding itself around AI

The Graviton deal is part of a larger shift in the company’s operations. The company is reducing traditional roles and investing in infrastructure for the future. CEO Mark Zuckerberg believes AI allows smaller teams to be more productive, influencing staffing decisions.

Meta is restructuring by merging teams, reducing management, and moving engineers to AI projects like Llama. They have also made smaller cuts in Reality Labs, hiring, and sales. US employees laid off in May will get 16 weeks of base pay plus two weeks per year of service and extended healthcare.

This trend is happening throughout the tech industry. Companies like Microsoft, Amazon, Oracle and Block are reducing their workforce while investing more in AI infrastructure.

The reason is consistent: they’re increasing investment in AI systems, carefully managing the number of employees, and redefining the roles that require human workers.

For Meta, the Graviton deal and the job cuts are two sides of the same strategy. The company is not contracting. It is redirecting, putting its resources where it believes the next decade of growth will come from.